Background
The BCL-2 family proteins regulate apoptosis primarily on the mitochondrial outer membrane through the intrinsic apoptotic pathway [
1,
2]. These proteins are divided into three classes based on their BCL-2 homology (BH) domains (BH1-BH4) and function [
3]: anti-apoptotic [BCL-2, BCL-xL, BCL-W, MCL-1, BCL2A1 (BFL-1, A1), and BCL-B], pro-apoptotic multi-domain effectors (BAX and BAK), and BH3-only proteins (e.g. BIM, PUMA, and NOXA). Inhibition of apoptosis is accomplished by sequestering pro-apoptotic proteins and thus preventing mitochondrial outer membrane permeabilization [
4].
Blocked apoptosis is a hallmark of treatment-resistant cancers and thus it suggests that BCL-2 family members have potential as clinical biomarkers [
1,
5]. In fact, several studies have linked BCL-2 family expression and response to chemotherapy in different types of cancers. It has been reported that patients with cancers highly “primed” to cross the apoptotic threshold exhibit superior clinical responses to chemotherapy [
6]. For chronic lymphocytic leukemia (CLL), high BCL-2 and MCL-1 expression levels have been reported to mediate resistance to chlorambucil, fludarabine, and rituximab [
7‐
9]. Although numerous studies have focused on the role of BCL-2 or MCL-1 in CLL, the role of other anti-apoptotic proteins and their contribution to clinical outcome is not clearly defined. Most importantly, pharmacologic inhibitors of BCL-2 family proteins are poised for widespread clinical use, so there is an immediate need for development of markers that can rationally direct and better personalize the use of these agents in the clinic [
10]. We have developed an anti-apoptotic BCL-2 family expression index that can predict the response of hematological cells, including CLL, as well as solid tumor malignancies, to the rationally designed BCL-2 family inhibitor, ABT-737/ABT-263 (navitoclax) [
11]. ABT-199 (venetoclax), a second-generation, rationally designed inhibitor that was re-engineered to bind selectively to BCL-2, shows anti-tumor activity in primary tumor cells and xenograft models [
12,
13]. Phase I clinical trials with ABT-199 have had high patient response rates that include many complete responses [
14].
Lymphoid malignancies, most commonly derived from B-cell precursors include more than 40 distinct tumor types, varying widely in phenotype and clinical behavior [
15]. CLL, the most common leukemia in the Western world [
16], is characterized by an expansion of small mature B cells in blood, lymph nodes, and bone marrow. Its heterogeneous clinical course [
17,
18] has led to a search for markers that can predict disease progression to allow better management of the disease. Mutational status of the immunoglobulin heavy chain variable (IGHV) region dichotomizes CLL patients into two risk categories: those with unmutated IGHV have an unfavorable prognosis, whereas patients with mutated IGHV tend to have a more favorable prognosis. ZAP70 and CD38 expression can serve as a surrogate for an unmutated IGHV gene, thus functioning as prognostic markers [
19,
20]. Despite of their clinical value, there are technical difficulties that preclude optimal use of these markers, such as standardization and reproducibility [
21,
22]. In addition, p53 deletion is a well-established marker of shorter survival and chemotherapy resistance [
23]; however, it is present in only a small percentage of patients with CLL at the initial diagnosis [
24,
25]. Overall, existing established prognostic markers fail to predict clinical outcome in a considerable number of patients with CLL [
24]. Moreover, it is difficult to integrate the results of these various markers to assess the overall risk in an individual patient [
22]. Thus, developing additional markers for CLL is of considerable interest as they may indicate inherent biologic differences that may be amenable to targeted therapeutic intervention.
MicroRNAs (miRs) are small non-coding regulatory RNAs that bind to a specific target mRNA through a sequence that is complementary primarily to the 3’-UTR of the target mRNA. They have roles in many underlying cancer processes, including proliferation, apoptosis, and invasion [
26,
27]. miRNAs are very stable and are found in body fluids such as plasma, serum, and urine, therefore cancer-specific miRNAs could potentially be used as a tumor molecular signature to track and predict cancer progression and to guide treatment [
28].
Here, we report that high BCL-xL expression inversely correlated with decreased levels of a newly identified miR-377. Mutational and functional analyses validated BCL-xL as a direct target of miR-377. Moreover, we show that BCL-xL/miR-377 regulation in diffuse large B-cell lymphoma (DLBCL) cells drives acquired therapeutic resistance to ABT-199 and is associated with advanced tumor stage in CLL patients. Collectively, these data support a model in which co-regulation of BCL-xL and miR-377 mediates a novel mechanism of acquired therapeutic resistance in B-cell lymphoid malignancies.
Discussion
CLL undergoes transformation (known as the Richter syndrome), to more aggressive lymphoma, most commonly DLBCL [
38]. In our GCB-DLBCL chemotherapeutic resistance model, chronic exposure to ABT-199 up-regulates BCL-xL expression, which is responsible for mediating chemotherapeutic resistance to ABT-199 [
31]. In search of the molecular mechanism that mediates high BCL-xL expression, we identified miR-377, expression of which was inversely correlated with that of BCL-xL. In our ABT-199 resistance model, we found that the high BCL-xL expression was due to down-regulation of miR-377, an observation similar to what we found in our CLL patients and a panel of lymphoid B-cell lines. We demonstrate that
BCL-xL is a direct target of miR-377 by three independent approaches: (i) a luciferase reporter assay, (ii) miR-377 expression modulation both by a mimic and an inhibitor and (iii) over expression of BCL-xL. While little is known about the molecular function of miR-377, in a comprehensive study using integrative genomic approaches, miR-377, among other miRNAs, was found to correlate with advanced tumor stage in solid tumors [
39,
40]. Here, we provide evidence for a tight regulation between miR-377 expression and up-regulation of BCL-xL. Based on these data, we propose a novel mechanism by which lymphoid malignant cells regulate BCL-xL and miR-377 expression in order to promote acquired resistance to chemotherapy. Interestingly, BCL-xL is one of the most frequently amplified oncogenes found in solid tumors [
41]. Here we identify a novel, miR-377-dependent regulation as an alternative mechanism for BCL-xL expression in leukemic cells.
miR-377 is located on chromosome 14q, with deletion of this site indicating its potential role as a tumor suppressor. 14q deletions are also associated with trisomy 12, a hallmark of CLL [
42]. As the frequency of the 14q deletion is relatively low in hematopoietic malignancies, including CLL [
33], it is not likely to provide a common therapeutic resistance mechanism. Interestingly, there is a growing interest in the 14q32 chromosomal region because ~ 53 miRNAs are embedded in two adjacent clusters (14q32.31 and 14q32.32); those span more than 200 kb, which were found to be deregulated in various human diseases, including various cancers, both hematologic, such as acute promyelocytic leukemia [
43] and of epithelial origin, such as melanoma [
44]. These reports support our findings that demethylating agents can restore expression of these miRNAs, including miR-377.
Our ABT-199R model suggests that an alternative, transcriptional silencing mechanism is responsible for the low levels of miR-377 that develop after chronic exposure to ABT-199. Our preliminary experiments using bisulfate genomic sequencing and methylation-specific PCR could not identify differentially methylated regions. Future experiments are needed to delineate how methylation regulates expression of miR-377 and other miRNAs in the 14q32 cluster. Nevertheless, the epigenetic modification of miRNA expression by methylation is well documented. Many miRNAs, including miR-377, are up-regulated following treatment with 5-Aza, a hypomethylating agent approved for use in myeloid, but not lymphoid malignancies [
36,
39,
45]. Treatment with 5-Aza led to re-expression of miR-377 leading to decreased BCL-xL expression. Interestingly, unlike parental cells, we found that ABT-199R cells were highly sensitive to 5-Aza. By modulating miR-377 expression with both a mimic and an inhibitor, along with re-expression of BCL-xL, we show that targeting BCL-xL by 5-Aza is a promising approach to overcome ABT-199 resistance. As several hypomethylating agents are currently in clinical use [
45,
46], this possibility could be quickly tested in clinical trials.
We show that CLL patients previously treated with a variety of chemotherapeutic regimens had higher BCL-xL and lower miR-377 expression as compared to untreated patients. Moreover, CLL patients with high BCL-xL/low miR-377 expression also have a more advanced tumor stage. These clinically relevant data along with our ABT-199R model suggest that BCL-xL might be critical for conferring general chemotherapy resistance. In support of our findings, other
in vitro studies also support a role for BCL-xL in promoting chemotherapy resistance. Thus, chemotherapeutic resistance to a group of compounds that repress MCL-1 expression has been linked to high
BCL-xL mRNA expression [
47]. This finding indicates that a patient-selection strategy for development of any MCL-1 inhibitor should focus on patients with low BCL-xL expression. Moreover, resistance to 122 standard chemotherapy agents correlated with high BCL-xL expression in the NCI 60 cell line panel [
48].
Although the role of BCL-2 and MCL-1 in the molecular pathogenesis of CLL has been extensively studied, less is known about the role of the other anti-apoptotic BCL-2 family members. Here, we have applied a highly sensitive and quantitative assay to examine the clinical value of expression for all anti-apoptotic BCL-2 family genes. Out of all these gene expression analyses, for individual or different combinations of these genes, only
BCL-xL expression had clinically relevant predictive value. In our patient cohort,
BCL-2 and
MCL-1 levels did not discriminate between those at risk for short treatment free survival and those who were not, consistent with previous reports [
8,
49].
In contrast to p53 (17p) deletions, ZAP70 and CD38 did not show significant correlations with treatment free survival in our patient population. Reports conflict on the significance of ZAP70 and CD38 in predicting clinical outcome in CLL, likely because of the lack of a standardized method for determining what constitutes positive and negative test results [
21,
22,
50]. Interestingly, we found that
BCL-xL expression levels could identify patients at high risk who were negative for ZAP70 or CD38, suggesting the possibility that it is a more robust and/or independent indicator of clinical outcome. While p53 deletions/mutations generally indicate aggressive disease with poor prognosis, the majority of CLL patients, which do not have this abnormality, remain heterogeneous.
BCL-xL expression levels were able to further stratify this group in terms of time to requiring treatment, indicating that
BCL-xL may also be useful in this context. Our findings that BCL-xL has prognostic value are not limited to CLL. Similarly, it was reported that high BCL-xL expression correlated with short overall survival in follicular lymphoma [
51]. As a further support for the role of BCL-xL in a wide range of B lymphoid malignancies, the clinical outcome of GCB-DLBCL patients is better than of those with non-germinal center-DLBCL, likely also due to the fact that GCB-DLBCL cells have significantly lower BCL-xL expression [
29,
52].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SA and AA participated in designing the research. SA performed the research and data analysis. GC developed and characterized ABT-199-resistant cells. JE and GS designed and performed the luciferase reporter assay experiments. BH and MS contributed the patient samples and participated in their characterization. VN and AT designed, performed and interpreted 14q32 cluster methylation experiments. TR provided the statistical analysis. SA and AA wrote the paper, and all authors reviewed the final version of the manuscript.